As basic building blocks, the equally ubiquitous amplifier and comparator can be designed into a host of circuits. The wide range of amplifier functions and large number of variants within each category indicate the high degree of specialization in this area. In many respects, circuit performance requirements can be used to specify almost everything about an amplifier except its part number. Unfortunately, few engineers can make the translations from systems and circuit parameters to equivalent IC specifications.
The drivers for changes in amplifiers are power consumption, packaging, precision, dynamic performance, and speed, not necessarily in that order. At the high-speed end, amplifiers will have bandwidths over 1 GHz and slew rates over 3000-V/µs. In the high-precision sector, look for all parameters approaching the ideal-zero for the detracting parameters and approaching theoretical limits for every other parameter.
In other amplifier sectors, the market is bifurcating to the low-cost discrete devices at the low end and performance at the high end. In some respect, the commodity parts are too low in cost to justify integrating into higher-level functions in a single chip.
For the lower-cost markets, amplifiers in standard CMOS with medium performance will be better matched to drive the integrated data converters and CPU devices. Many devices are moving toward CMOS for its cost advantages over other processing technologies. Another advantage of CMOS is its ability to integrate more digital techniques, such as choppers into an analog circuit. Incorporating digital functions for device error corrections reduces the need to process the data in the DSP.
The proliferation of specialized amplifier types to some extent has completely displaced the generic op amp. Designers must address the selection and definition of the appropriate specifications. This cornucopia of parts ensures that a designer can pick from a set of mutually exclusive specifications and make a tradeoff between partial matches of those key requirements.
>FOR THE GREATEST PERFORMANCE, discrete analog function blocks will still be the best way to implement a design. The additional costs of a mixed-signal module to a standard CMOS process doesn't justify the replacement of a few sub-$1 parts in the systems. In addition, the performance of the discretes is usually much higher than the integrated functions.
>SMALL PACKAGES WILL EMERGE for ever more integrated functions. The ultimate size reduction is to a chip-scale package, making the board penalty for separate analog functions almost nonexistent, except for the bypass capacitors. The incorporation of more functions in a single package improves thermal matching for the subsystem.
>NEW DESIGNS MUST OPERATE over wider supply ranges. In some cases, the parts will operate from under 1.5 V up to ±15 V. The ability to operate on a wide range of supplies reduces the number of amplifier variants needed in the product mix. The decreasing digital supplies work against noise immunity and analog accuracy, so a separate, higher analog supply helps the overall system error budget.
>LOWER-OFFSET, LOWER-INPUT BIAS CURRENTS and higher-input impedance will spur on higher performance. By moving toward 16-bit accuracy, parts will require lower short-term and long-term drift and smaller parameter shifts due to temperature changes. All integrated amplifiers will have lower distortion and higher linearity to get their performance to at least a 16-bit level or higher. The error budget will be reserved for the sensors.
>FOR CMOS DEVICES, low parasitic ESD up to ±15 kV will become a necessity as the outside world comes into contact with sensors and interfaces. The amplifiers are too close to the input devices to be exempt from the ESD but cannot tolerate the loading that the digital protection circuitry imposes.
More designs will employ complementary devices. A complementary process is better suited for balanced drive and faster parts. The decrease in supplies requires dramatic changes in architecture to retain the necessary dynamic range and operating headroom for the existing applications. In addition, the amplifiers must address the changing architectures of the data converters, and the complementary structures enhance architectural creativity.
>FIELD-PROGRAMMABLE ANALOG ARRAYS should start to take market share from the low-end amplifiers and associated components for functions like filters and analog conditioning blocks. Not only do the programmable analog arrays bring the building blocks together, they often are reprogrammable. The ability to configure an analog building block in-circuit eases the design tasks and lets a digital engineer of programmer do the first pass on the analog subsystem.
>SOFTWARE POTENTIOMETERS: Programmable individual components will continue to be viable in reducing component counts and system-level design complexity. The ability to change device parameters in-circuit enables designers to reduce the total amount of devices and ease manufacturing. In addition, components with programmable parameters can take advantage of the vast amounts of digital processing available in the systems while reducing the latency and processing overhead in the DSP sections.
>THE NEED FOR BETTER MATCHING of functions and specifications to the end application will force vendors to create more special functions parts. For example, if customers want an ultra-low-voltage amplifier with very low output impedance, they'll choose application-specific products that offer the appropriate combination of low voltage and low output impedance specifications. Such products may be described in the data sheets as an ADSL amplifier, a weigh-scale amplifier, or a taillight sensor amplifier.
>SUPERSPEED AMPLIFIERS WILL APPEAR WITH bandwidths of greater than 2 GHz. Their gain-bandwidth products will exceed 10 GHz. The challenge for the users will be in creating a stable amplifier that operates in the microwave frequencies. The ultra-high-speed amplifiers will force changes in topologies and architectures in any application that can use an amplifier that is not a 50-W RF module.
Ultra-precision amplifiers will also move up in frequency response to address the improvements in data converter capabilities. They also will step up to capture those small, high-speed signals that are not accessible with current precision amplifiers. Expect to see applications such as sensors that are approaching the parts-per-trillion range. In turn, these sensors will be used in physics and chemistry.